Wheeler “Prima Facie” Necessity page 1 11 22 2013

“Prima Facie” Necessity: Introduction to an Adequate Account of “Ought”

Section I: Doxa

Consider the following two remarks, and suppose they are both true (as they sort of seem to be):

1) “If Fred is an NRA member [N], he’s gotta be a conservative[C].”

2) “If Fred is for legalized recreational drugs[R], he can’t be a conservative[C].”

Suppose, as is possible, that Fred is both an NRA member and for legalizing drugs. Then these modal conditionals are true even though the de-modalized conditionals are not.[1] That is, taking the truth-functional conditional to be the minimal truth-condition for any conditional, and assuming that if something is necessary then its true, we get [(N/\R) /\ (-N v C) /\ (-R v -C)], which is contradictory.

A theory that allows 1) and 2) to be simultaneously true is the following: “Necessary” and its synonyms “must be” “gotta be” and the like, have a “prima facie” sense as well as an “absolute” sense. A conditional can be prima facie necessary without its truth-functional correspondent being true. The theory is that some senses of necessity are “prima facie” necessities.

What does “prima facie” mean? It means that there is some connection between the states of affairs mentioned in the antecedent and the consequent, and that other things being equal, that connection is manifest and operative. Thus, prima facie necessity is equivalent to the “real connection” sense of “probably.” “If Fred is an NRA member, he is probably a conservative” is an alternative to the “prima facie” use of necessity. So “probably,” another “connection” term, is assimilated to necessity.

The key to this theory is the realization that, in talking about necessity, we aren’t doing semantics or logic, but rather giving an account of metaphysical connection, a property that comes in degrees.

Many objections could be raised against this theory, along the following lines:

a) “Sentences 1) and 2) cannot both be true.”

But that is because the person is tied to the “absolute” understanding of necessitation and does not grasp the “prima facie” understanding of necessitation. Some necessities obtain “other things being equal.”

b) “The `logic’ of necessity rules out this theory.”

But necessity is a feature of the world and real connection, not a matter of logic and semantics. The important thing is that we access the truths about necessity and probability, not that we get “semantics” “right” by paying attention to linguistic irrelevancies.

c) “When we use “probably” in talking about lottery chances or the likelihood that a finesse will succeed, necessary connection is not an issue.”

That just shows that there are many senses of “probably.” The non-metaphysical senses are not at issue here, and are not relevant to getting an account of the truths of real connection.

This is clearly a teflon theory. For any apparent counter-example, we can either appeal to different senses of “necessity,” and “probability,” excluding counterexamples as irrelevant, or we can invoke the “prima facie” argument.

II Alethia: Critique and an alternative

I think you’ll agree this theory is really crummy. The cost of accommodating 1) and 2) is too high, and those sentences should be taken to be hyperbolic. But what’s really wrong with the theory? The theory is insulated from any considerations that could be brought against it. A theory-format in which every problem can be overcome by postulating new “senses” of terms is unfalsifiable and unproductive, since no hypotheses are forthcoming. Yet a precisely analogous theory seized the minds of ethical thinkers quite a while ago. The logic of “ought” was treated as the logic of obligation, which indeed behaves like (traditional) necessity. The unhelpful fudge of “prima facie obligation” was supposed to fit such a theory to some intuitive facts.[2]

To begin to have a theory of a subject matter, you have to have a theory of how the inferences go through, which amounts to a theory of what the sentences say, what their logical form is. So consider how we would approach a theory of “if A then probably B” sentences in contrast to the approach to “if A then necessarily B” sentences: “Probably” is primarily epistemic, a characterization of what is believable given the evidence; whereas “necessarily” is ontological. “Probably” is supported by reasons to believe that something is the case, and governed by norms of what it is rational to believe, given other beliefs. The intuitive reason probability-sentences do not detach absolutely probable consequents from the truth of their antecedents is that there can be good reasons both for and against a claim. “Necessity,” on the other hand, necessitates. Reasons making something necessary are not norms for rational beliefs, but laws of nature or something else that guarantees the truth of the consequent from the truth of the antecedent. Of course, if you know that something is a law of nature, that’s an epistemic reason to believe it. Reasons making something necessary are not counter-balanced.[3]

Roughly, “probably” is the modality for inductive inference, “necessity” for deductive inference. The most important feature of inductive argument is the following: A strong inductive argument can be weakened by the addition of further information. You can have very compelling evidence that Fred is the murderer, only to have it overwhelmed by a newly-discovered videotape. On the other hand, if a deductive argument is valid, then any additional information cannot affect its validity.

A central thesis of Davidson’s “How is Weakness of the Will Possible” is that ethical reasoning is fundamentally like induction rather than deduction. Here is an application of the “inductive” model: The inductive/deductive distinction corresponds to one aspect of Gilligan’s “justice” versus “care” theories [perspectives], for instance. The basic idea of the “obligation” way of thinking about ethics is deductive and matches the features of what Gilligan has called the justice perspective. But, if ethical reasoning is like induction, then ethical skill doesn’t tell you what to do in “story-problem” situations. So, Kohlberg’s “ethical dilemmas” are bogus, and the women were right to reject them. Given the conditions of the story-problem dilemma, the answer could be “yes” or “no” depending on other surrounding circumstances. An ethically sensitive person will not feel in a position to judge, because obviously discoverable additional information could always change the judgment. More generally, anyone attracted to the idea that ethical judgment is contextual should favor an inductive account that makes sense of the idea. So the distinction between "ought" and "obligation" is by no means trivial, as our job candidates this year supposed: it makes all the difference in the very way ethics is conceived.

The hypothesis is that “ought” is a modal connective or predicate formally akin to “probably.” “Ought”-conditionals do not detach, because they generally depend on inductive-like, non-conclusive reasons for belief, action, and desire. “Ought” conditionals express that modal connection between the antecedent and the consequent. I will sketch the theory of such connection below. For the moment, let us talk about “senses” generally and then the “senses” of that modal connection:

a) Senses

The main reason philosophers don’t want to take a semantics for “ought” seriously is that there are clearly some “ought” sentences that have nothing to do with ethics or human action. “If you turn the key, your car ought to start” has little to do with the great traditions of ethics. So, obviously, there is a different non-moral “sense” of “ought” occurring. This objection reveals too casual thinking about senses, and a legitimate concern about the source of normativity. To discuss the senses of “sense” first, observe that if “ought” is a modality, there are lots of possibilities for it being the same modality wherever it occurs.

Consider a trivial case, the senses of “tall:” The Gugenheim is not tall; the Empire State Building is. Yet the Gugenheim is much taller than the very tall Michael Jordan. A theorist could say the following: There are many senses of tall, corresponding to the various categories of objects said to be tall. There is a sense for human beings, a sense for buildings, and so on.[4] The motivation for having different senses would be the observation that, if “tall” is univocal, then if A is tall and B is taller than A, then B is tall. Where this breaks down, as with Jordan and the Gugenheim, another sense of “tall” must be invoked to remove the paradox.

One could propose logically unrelated predicates for buildings and people. (Perhaps the justification would be that different areas of the brain process persons and buildings.) But every respectable[5] theory of the “senses” of tall interprets different senses as relativizations-- to a class, for instance, or to a context-determined parameter.[6] So, the “logical form” of “Jordan is tall” would contain an implicit relativization. “Jordan is tall” would be a contextually abbreviated form of “Jordan is a tall man,” for instance.[7]

So: Different senses: Different senses may be:

1) different predicates: can, can; bank, bank; swim, swim; bore, bore, bore

2) the same predicate but with different relativizations

b) senses of “ought”

The obvious working hypothesis is that “ought” is a modality, a relation among whatever you would like “that” clauses to designate. A beginning hypothesis is that it is something akin to probability: The “consequent” has some connection with what it is reasonable to believe relative to the “antecedent.” How can this relation apply both to “If the key is turned, the car ought to start” and to “Fred ought not to stomp on Bill?”

A beginning is to notice that the “key is turned/car starts” “ought”-sentence is usually true when the corresponding probability sentence is true. So, where can we find “probability” claims that would play corresponding roles in “ought”-sentences about what human beings ought to do in various situations? Step one is to realize that the constraints on interpretation that Quine and Davidson take to be fundamental to understanding rational behavior are probabilistic. That is, they are features to be maximized in interpreting the behavior of another as reasonable. Therefore, they amount to probabilistic laws about human beings. Alternatively, they are “ceteris paribus” laws about rational agents.

Step two is to hypothesize that “ought” sentences are true just in case there is a chain of applications of such laws such that the first element of the chain is the “antecedent” of the “ought” conditional and the last element is the consequent. Roughly speaking, what ought to happen what would happen if what probably was going to happen always happened. So, for instance, the result one ought to arrive at in a long calculation is the result one would arrive at if one always calculated as one usually calculates, for the individual steps. (One nice feature of the theory is that it predicts the right results for very-short chains—they will turn out to be identical with conditional probability chains, as in the car-starting example.)

That is, one way to get a theory of “ought” from the resemblance between “probable” and “ought” is that chains of such maximized features tell you what a person ought to do, relative to a set of features to maximize. A case where this clearly works is the “mathematical” ought, which can be summarized as “you ought to believe the mathematical consequences of what you believe.” For complex reasoning, for most of us, this rarely happens, although it is also rare that we make an elementary mistake. The idea is that the answer you ought to get is the answer you would get if you always multiplied the way you usually multiply. Complex inferences are sequences of small inferences most of which are accurate. So, you ought to get the right answer, even though you probably won’t, and even though you probably will get the right answer at each step. (multiplication of probabilities.)

The theory supposes several things about rationality and about morality:

1) First, it supposes that “principles of rationality,” as empirical accounts of rational agents, are all of the “if…then…probably” type.

2) Second, it supposes that a complex situation about what a person ought to do can be broken down into smaller steps; that there is something analogous to a calculation involved in deciding what to do.

3) Third, it supposes that there are various kinds of considerations that can yield more or less limited “ought”-conclusions about what persons should do.

Let me begin by developing this third aspect: Our understanding of humans is much more complex than our understanding of cars, since many different kinds of considerations affect their actions. Relative to some kinds of considerations, certain actions are reasonable in certain circumstances, even though relative to other kinds of considerations, an incompatible action might be reasonable in those same circumstances.

Humans are both biological entities and purposefully choosing entities. If I’ve slipped Fred some LSD, I can say, “He ought to start hallucinating about now,” even though he’s just beginning his final public oral, and, “in another sense” ought to be in command of his information links to the outside world.

Furthermore, within the category “purposefully choosing entities” there are different kinds of grounds. Start with a simple example: Fred is playing chess with Celeste, and there is a nice f7 bishop sacrifice that leads to mate. Other moves leave it an even game. Clearly, if the mate is really forced, Fred should sacrifice at f7.

Celeste, however, is Fred’s boss. So, if Celeste is a poor and vengeful sport, Fred should not sacrifice at f7. Or, if the bishop being on f7 will complete a circuit that disrupts Fred’s pacemaker, Fred should not sacrifice at f7. Or, if Celeste has problems with self-esteem and is only being sustained in sanity by a mistaken belief in her chess prowess, Fred should not sacrifice at f7. (This is a good Gilligan example: if Celeste, having moved, also sees the f7 sac, what ought you to do then? It depends. There is no general rule for induction, and no general rule telling you what to do in advance of a concrete situation.)

Relative to different kinds of considerations (the chessic, the prudential, the sympathetic) Fred ought to do different things. Each area of consideration, furthermore, will have its “ought”-principles about how these considerations bear on questions of action. So, the same modal connection can hold between normative considerations and others. Now, what about these “considerations.”

c) normativity

The fundamental objection people have to the assimilation of moral “oughts” to other “senses” of “ought” is that the moral oughts have normative force. Does this theory imagine that normativity is somehow reduced to a common modality? No way. Normativity, according to Davidson and Sellars, comes in already with the ascription of beliefs and desires. So this theory-format is by no means an attempt to derive normativity from something else. The fact that the well-understood “logical ought” is essentially normative and is required in interpretation, that is, in the ascription of beliefs and desires, is an argument that beliefs and desires are not going to turn out to be physically definable states of organisms, where “physical” means “not structured by rational norms.” So, given that ascribing beliefs and desires to a person maximizes the degree to which they adhere to norms, an “ought” essentially describing perfection (as we will see below) in adherence to these norms will be normative.

So where do the “sets of considerations” come from? My hypothesis is that all such sets of considerations are parts of reason, either reason relative to certain ends or reason tout court. “Reason” is “what one should, other things being equal, maximize in understanding action.” Thus reason includes reasonable contents of desires, appropriate emotional response, reasonable hypothesis-selection procedures, reasonable inductive inferences, and the like. Roughly, what a person should want, feel, believe and infer, relative only to the information that he is a person.[8] It is provable that “reason” is not axiomatizable. For one thing, it is provable that so-called inductive logic is not axiomatizable.

Conclusion:

1) abandon “prima facie” obligation

2) recognize that appeal to “senses” is not giving an answer

3) distinguish “ought” from “obligation” and make “ought” the primary ethical modality[9]

[1] Thus this phenomenon is different from the well-known megarian argument. There, you have sentences such as the following:

1a) If Fred has siblings, he’s gotta have brothers or sisters. N(S ->B)

2b) If Fred is an only child, he can’t have brothers or sisters. N(-S-> -B)

From 1a), the megarian fallacy is to conclude that, given that Fred has siblings, that Fred was doomed to the misery of sharing his parents’ attention, and that more generally, whatever is the case must be the case.

In this case, the “paradox” is resolved by the notion of the scope of a modal operator, simply distinguishing between the necessity of the conditional and a conditional with a necessity as consequent.

(Contrast: If P is a tautology, then P is necessarily true)

[2] The reason for the assimilation may have had to do with a theological account that made a very strong connection between what one ought to do (as a rational being) and what one is obligated, or legally bound, to do. If indeed one achieved salvation by following the commands of God, then what one ought to do would match what one was obliged to do. God’s laws had the structure of a deductive system, at least in principle (the Talmud is proof that such systems are incomplete). With the end of monotheism as a presupposition of philosophy, the idea that what one ought to do was what was obligatory “by moral principles” continued.